Local learning in local model networks
Roderick Murray‐Smith · 1995
Local Model Networks are hybrid models which allow the easy integration of a priori knowledge, as well as the ability to learn from data to represent complex, multidimensional dynamic systems from data. This paper points out problems with global learning methods in Local Model Networks. The bias/variance trade-offs for local and global learning are examined, and it is illustrated that local learning has a regularizing effect that can make it favorable compared to global learning in some cases. 1 Local Model Networks The basic assumption underlying the use of learning systems for modelling purposes is that the behaviour of the system can be described in terms of a training set DN = ((/(1); y(1)); :::; (/(N ); y(N ))) consisting of its observed input vector / and corresponding scalar output y. We assume the system output can therefore be modelled as y = f(/) + ": (1) where f is a function, and " is independent random measurement noise with zero mean and variance oe 2 . The modelling ...